Carbonyl Reductases and Pluripotent Hydroxysteroid Dehydrogenases of the Short-Chain Dehydrogenase/Reductase Superfamily

Hoffmann, Frank

Carbonyl Reductases and Pluripotent Hydroxysteroid Dehydrogenases of the Short-Chain Dehydrogenase/Reductase Superfamily

Structural Aspects of Oligomerization in 3-Hydroxysteroid Dehydrogenase/Carbonyl Reductase from Comamonas testosteroni

von Frank Hoffmann (Autor:in)

Chemie - Biochemie

Zusammenfassung

Inhaltsangabe:Introduction:
Metabolic reduction is the counterpart to oxidative pathways and plays an important role in the phase-I metabolism of carbonyl group bearing substances. Carbonyl reduction means the formation of a hydroxy group from a reactive aldehyde or ketone moiety and is generally regarded as an inactivation or detoxification step since the resulting alcohol is easier to conjugate and to eliminate. Not only are these carbonyl-containing compounds widespread in the environment and enter the body as xenobiotics and environmental pollutants, but they can also be generated endogenously through normal catabolic oxidation and deamination reactions. Many endogenous compounds such as biogenic amines, steroids, prostaglandins and other hormones are metabolized through carbonyl intermediates. In addition, lipid peroxidation within the cell results in the production of reactive carbonyls such as acrolein, 4-hydroxynonenal, 4-oxononenal and malon-dialdehyde, while oxidative damage to DNA generates base propenals. Dietary sources of carbonyl-containing compounds are diverse and include aldehydes found in fruits as well as the breakdown product of ethanol, acetaldehyde. Pharmacologic drugs represent further sources of exposure to carbonyl-containing compounds.
From the pharmacologists point of view, carbonyl reduction has been shown to be of significance in various inactivation processes of drugs bearing a carbonyl group. On the other hand, the carbinols formed may retain therapeutic potency, thus prolonging the pharmacodynamic effect of the parent drug, or, in some instances, a compound gains activity through carbonyl reduction.
From the toxicologists point of view, carbonyl reduction plays an important role in the toxification of drugs such as daunorubicin and doxorubicin (cf. chapter 4), whereas numerous reports corroborate the concept of carbonyl-reducing enzymes being involved in detoxification processes of endogenous and xenobiotic reactive carbonyl compounds.
Compared with the oxidative cytochrome P450 (CYP) system, carbonyl-reducing enzymes had, for a long time, received considerably less attention. However, the advancement of carbonyl reductase molecular biology has allowed the identification and characterization of several carbonyl-reducing enzymes, including pluripotent hydroxysteroid dehydrogenases that are involved in xenobiotic carbonyl compound metabolism, in addition to catalyzing the oxidoreduction of their physiologic […]

Leseprobe

Inhaltsverzeichnis

Frank Hoffmann

Carbonyl Reductases and Pluripotent Hydroxysteroid Dehydrogenases of the Short-

Chain Dehydrogenase/Reductase Superfamily

Structural Aspects of Oligomerization in 3-Hydroxysteroid Dehydrogenase/Carbonyl

Reductase from Comamonas testosteroni

ISBN: 978-3-8366-3391-8

Herstellung: Diplomica® Verlag GmbH, Hamburg, 2009

Zugl. Philipps-Universität Marburg, Marburg, Deutschland, Dissertation / Doktorarbeit,

2009

Dieses Werk ist urheberrechtlich geschützt. Die dadurch begründeten Rechte,

insbesondere die der Übersetzung, des Nachdrucks, des Vortrags, der Entnahme von

Abbildungen und Tabellen, der Funksendung, der Mikroverfilmung oder der

Vervielfältigung auf anderen Wegen und der Speicherung in Datenverarbeitungsanlagen,

bleiben, auch bei nur auszugsweiser Verwertung, vorbehalten. Eine Vervielfältigung

dieses Werkes oder von Teilen dieses Werkes ist auch im Einzelfall nur in den Grenzen

der gesetzlichen Bestimmungen des Urheberrechtsgesetzes der Bundesrepublik

Deutschland in der jeweils geltenden Fassung zulässig. Sie ist grundsätzlich

vergütungspflichtig. Zuwiderhandlungen unterliegen den Strafbestimmungen des

Urheberrechtes.

Die Wiedergabe von Gebrauchsnamen, Handelsnamen, Warenbezeichnungen usw. in

diesem Werk berechtigt auch ohne besondere Kennzeichnung nicht zu der Annahme,

dass solche Namen im Sinne der Warenzeichen- und Markenschutz-Gesetzgebung als frei

zu betrachten wären und daher von jedermann benutzt werden dürften.

Die Informationen in diesem Werk wurden mit Sorgfalt erarbeitet. Dennoch können

Fehler nicht vollständig ausgeschlossen werden und der Verlag, die Autoren oder

Übersetzer übernehmen keine juristische Verantwortung oder irgendeine Haftung für evtl.

verbliebene fehlerhafte Angaben und deren Folgen.

http://www.diplomica.de, Hamburg 2009

Part I

Carbonyl Reductases and Pluripotent Hydroxysteroid Dehydrogenases of

the Short-Chain Dehydrogenase/Reductase Superfamily

Page 2 ff

Part II

Structural Aspects of Oligomerisation in 3-Hydroxysteroid Dehydro-

genase /Carbonyl Reductase from Comamonas testosteroni:

New Approaches for efficient Protein Design

Page 69 ff

Part I

Carbonyl Reductases and Pluripotent Hydroxysteroid Dehydrogenases of

the Short-Chain Dehydrogenase/Reductase Superfamily

Frank Hoffmann

Institute of Toxicology and Pharmacology for Natural Scientists,

University Medical School Schleswig-Holstein, Campus Kiel,

Brunswiker Strasse 10, 24105 Kiel, Germany

Published in:

Drug Metabolism Reviews

Volume 39, Issue 1 January 2007

Pages 87 144

Table of Contents

Part I

Table of Contents

Abstract ... 5

Introduction ... 6

I.1 Carbonyl

Reduction ... 6

I.2

Enzymes Mediating Carbonyl Reduction ... 7

1.1

The Aldo-Keto Reductase (AKR) Superfamily... 9

1.2

Short-Chain Dehydrogenase/Reductase (SDR) Superfamily ... 10

I.3

Hydroxysteroid Dehydrogenases as Carbonyl Reductases... 11

General Features of the SDR Superfamily Enzymes... 13

II.1 Historical

Background:

Functional Characterization ... 13

II.2

SDR Superfamily Classification... 14

II.3

Structural Features of the SDR Members ... 16

II.3.1 Catalytic Triade and Catalytic Mechanism ... 18

II.3.2 Substrate Binding and Substrate-Binding Loop ... 21

II.3.3 Cofactor

Binding ... 22

II.3.4 C-terminal Extension and 3

-Helices ... 23

II.3.5 Oligomerization

and

Interfaces ... 23

III

Pluripotent Carbonyl Reductases of the SDR Superfamily... 27

III.1 Carbonyl Reductases in Non-Mammals ... 29

III.1.1 3

/20-Hydroxysteroid Dehydrogenase of Streptomyces hydrogenans... 29

III.1.2 3

-Hydroxysteroid Dehydrogenase/Carbonyl Reductase of Comamonas

testosteroni... 30

III.1.3 Insect Carbonyl Reductase: Sniffer of Drosophila melanogaster ... 31

III.2 Carbonyl Reductases in Mammals ... 32

III.2.1 Monomeric Cytosolic NADPH-Dependent Carbonyl Reductases... 32

III.2.1.1 Human Carbonyl Reductase 1 (CBR1) ... 32

III.2.1.2 9-Keto-Prostaglandin Reductase and 15-Hydroxy-Prostaglandin

Dehydrogenase... 36

III.2.1.3 Human Carbonyl Reductase 3 (CBR3) and 4 (CBR4) ... 37

Table of Contents

III.2.1.4 Chinese Hamster Carbonyl Reductases (CHCR 1-3)... 38

III.2.1.5 Rat Carbonyl Reductases (iCR, nCR, rtCR)... 39

III.2.1.6 Pig Testicular Carbonyl Reductase (PTCR) ... 40

III.2.1.7 Tetrameric Peroxisomal Carbonyl Reductase ... 41

III.2.1.8 Tetrameric Mitochondrial Carbonyl Reductases ... 41

III.2.1.9 Dimeric Microsomal Carbonyl Reductase: 11

-HSD Type 1... 43

Biological Functions of Carbonyl-Reducing Enzymes ... 48

IV.1 Roles in Steroid and Prostaglandin Metabolism... 48

IV.2 Tetrahydrobiopterin

Synthesis... 49

IV.3 Neuroprotection by Carbonyl Reductase? ... 51

IV.4 Quinone

Detoxification... 54

IV.5 Carbonyl Reduction in Drug Metabolism and Pharmacology... 56

IV.6 Role in Chemotherapy Resistance ... 58

IV.7 Protection against Tobacco Smoke-Derived Lung Cancer... 61

IV.8 Detoxification of Insecticides ... 65

Physiological Implications... 66

Perspectives... 67

Zusammenfassung... 68

References ... 161

VII

Figure/Table List:... 210

Abstract

I Abstract

Carbonyl reduction of aldehydes, ketones and quinones to their corresponding hydroxy de-

rivatives plays an important role in the phase-I metabolism of many endogenous (biogenic

aldehydes, steroids, prostaglandins, reactive lipid peroxidation products) and xenobiotic

(pharmacologic drugs, carcinogens, toxicants) compounds. Carbonyl-reducing enzymes are

grouped into two large protein superfamilies, the aldo-keto reductases (AKR) and the short-

chain dehydrogenases/reductases (SDR). Whereas aldehyde reductase and aldose reductase

are AKRs, several forms of carbonyl reductase belong to the SDRs. In addition, there exist a

variety of pluripotent hydroxysteroid dehydrogenases (HSDs) of both superfamilies which

specifically catalyze the oxidoreduction at different positions of the steroid nucleus, and

which also catalyze rather non-specifically the reductive metabolism of a great number of

non-steroidal carbonyl compounds. The present review summarizes recent findings on car-

bonyl reductases and pluripotent HSDs of the SDR protein superfamily.

Introduction

II Introduction

II.1

Carbonyl Reduction

Metabolic reduction is the counterpart to oxidative pathways and plays an important role in

the phase-I metabolism of carbonyl group bearing substances. Carbonyl reduction means the

formation of a hydroxy group from a reactive aldehyde or ketone moiety and is generally re-

garded as an inactivation or detoxification step since the resulting alcohol is easier to conju-

gate and to eliminate. Not only are these carbonyl-containing compounds widespread in the

environment and enter the body as xenobiotics and environmental pollutants, but they can

also be generated endogenously through normal catabolic oxidation and deamination reac-

tions. Many endogenous compounds such as biogenic amines, steroids, prostaglandins and

other hormones are metabolized through carbonyl intermediates (Felsted and Bachur, 1980;

Forrest and Gonzalez, 2000). In addition, lipid peroxidation within the cell results in the pro-

duction of reactive carbonyls such as acrolein, 4-hydroxynonenal, 4-oxononenal and malon-

dialdehyde, while oxidative damage to DNA generates base propenals (Esterbauer et al.,

1982). Dietary sources of carbonyl-containing compounds are diverse and include aldehydes

found in fruits as well as the breakdown product of ethanol, acetaldehyde (Ellis and Hayes,

1995; Maser and Oppermann, 1997). Pharmacologic drugs represent further sources of expo-

sure to carbonyl-containing compounds (Maser, 1995; Rosemond and Walsh, 2004) (cf. chap-

ter 4.5).

From the pharmacologist's point of view, carbonyl reduction has been shown to be of signifi-

cance in various inactivation processes of drugs bearing a carbonyl group. On the other hand,

the carbinols formed may retain therapeutic potency, thus prolonging the pharmacodynamic

effect of the parent drug, or, in some instances, a compound gains activity through carbonyl

reduction.

From the toxicologist's point of view, carbonyl reduction plays an important role in the toxi-

fication of drugs such as daunorubicin and doxorubicin (cf. chapter 4), whereas numerous

reports corroborate the concept of carbonyl-reducing enzymes being involved in detoxifica-

tion processes of endogenous and xenobiotic reactive carbonyl compounds.

Introduction

Compared with the oxidative cytochrome P450 (CYP) system, carbonyl-reducing enzymes

had, for a long time, received considerably less attention. However, the advancement of car-

bonyl reductase molecular biology has allowed the identification and characterization of sev-

eral carbonyl-reducing enzymes, including pluripotent hydroxysteroid dehydrogenases that

are involved in xenobiotic carbonyl compound metabolism, in addition to catalyzing the oxi-

doreduction of their physiologic steroid substrates [reviewed in (Maser, 1995)].

II.2

Enzymes Mediating Carbonyl Reduction

Enzymes catalyzing the reductive metabolism of carbonyl compounds are ubiquitous in na-

ture, but the majority of the research work was performed in mammalian cells and tissues.

Carbonyl-reducing enzymes have been reported in a variety of non-mammalian organisms,

such as plants (Baker and Agard, 1994), bacteria (Zelinski et al., 1994; Oppermann and Ma-

ser, 1996; Oppermann et al., 1998), yeast (Peters et al., 1993; Wada et al., 1998), fish and

insects (Oppermann et al., 1998). In mammals, these enzymes have been analyzed in more

detail, amongst others, in rabbit (Ahmed et al., 1978; Ahmed et al., 1979; Sawada et al.,

1979; Felsted and Bachur, 1980; Imamura et al., 1993), rat (Ahmed et al., 1978; Hara et al.,

1987), mouse (Ahmed et al., 1978; Nakayama et al., 1986; Iwata et al., 1990a), guinea pig

(Sawada et al., 1979; Nakayama et al., 1986), monkey (Lee and Levine, 1974a; Iwata et al.,

1990a), chicken (Lee and Levine, 1974b; Cagen et al., 1979), pig (Tanaka et al., 1992), dog

(Hara et al., 1986b; Iwata et al., 1990a) and bovine species (Iwata et al., 1990a).

Human carbonyl-reducing enzymes occur in many different tissues like liver (Ahmed et al.,

1979; Sawada and Hara, 1979; Sawada et al., 1979; Felsted and Bachur, 1980; Hara et al.,

1986b), lung (Nakayama et al., 1986; Nakanishi et al., 1995), heart (Imamura et al., 1999a;

Imamura et al., 1999b), kidney (Higuchi et al., 1993; Imamura et al., 1993), brain (Wermuth,

1981; Cromlish et al., 1985; Iwata et al., 1989; Iwata et al., 1990b; Inazu et al., 1992b;

Wermuth et al., 1995), ovary (Iwata et al., 1989; Aoki et al., 1997; Guan et al., 1999), and

adrenal (Oppermann et al., 1991; Maser et al., 1992) and are known to exhibit broad substrate

specificities for xenobiotic aldehydes and ketones as well as for physiological carbonyl com-

pounds such as biogenic aldehydes (Smolen and Anderson, 1976), prostaglandins (Hayashi et

al., 1989; Inazu and Satoh, 1994), and steroid hormones (Pietruszko and Chen, 1976; Ikeda et

al., 1984; Maser, 1995).

Introduction

The difference in tissue and intracellular distribution suggests that several enzymes may be

involved in the enzymatic reduction of carbonyl compounds and that the intracellular multi-

plicity of the enzymes may have some relation to their physiological function. Since the early

seventies, a multitude of carbonyl-reducing enzymes has been purified and characterized in

terms of substrate specificity and kinetic constants (Fig.1). While lots of data on these en-

zymes were accumulating, it was speculated that several of these enzymes are in fact identical

to one another. On the basis of their primary structure, three of these enzymes were classified

as aldehyde reductase (EC 1.1.1.2), aldose reductase (EC 1.1.1.21) (Bohren et al., 1989) and

carbonyl reductase (EC 1.1.1.184) (Wermuth et al., 1988) (Fig. II-1) . As proposed by the

scientific community working in this field, these enzyme names were adopted in the Interna-

tional Enzyme Nomenclature System and received their respective EC number. In addition, it

turned out that there exist several isoforms of dihydrodiol dehydrogenase, which, initially,

were summarized under EC 1.3.1.20. An interesting observation was that several hydroxys-

teroid dehydrogenases (HSDs) specifically catalyze the oxidoreduction at different positions

of the steroid nucleus on the one hand, but on the other hand rather non-specifically catalyze

the reductive metabolism of a great variety of non-steroidal xenobiotic carbonyl compounds.

This was also true for the dihydrodiol dehydrogenases which were later shown to be HSDs

(cf. chapter 1.2.2 of this review) (Ratnam et al., 1999).

alcohol: NADP

-oxidoreductase

high Km aldehyde reductase

mevaldate reductase

daunorubicin-pH 8.5 reductase

hexonate dehydrogenase

lactaldehyde reductase

glucuronate reductase

ALR 1

alditol: NADP

-oxidoreductase

low Km aldehyde reductase

ALR 2

NADPH: sec.-alcohol oxidoreductase

prostaglandin 9-ketoreductase

daunorubicin-pH 6.0 reductase

ALR 3

several isoforms of dihydrodiol

dehydrogenases and/or

several HSDs specified according to

carbon atom involved in catalysis

aldehyde reductase

(EC 1.1.1.2)

carbonyl reductase

(EC 1.1.1.184)

hydroxysteroid dehydrogenases

(AKR1C family)

hydroxysteroid dehydrogenases

(EC 1.1.1.xx)

aldose reductase

(EC 1.1.1.21)

AKR

SDR

Fig. II-1: Multiplicity of enzymes involved in carbonyl reduction

Introduction

Sequence comparisons and data base searches revealed that the enzymes mediating carbonyl

reduction obviously belong to two large protein superfamilies. Whereas aldehyde reductase

and aldose reductase belong to the aldo-keto reductases (AKR) (Jez et al., 1997a), carbonyl

reductase turned out to be a member of the short-chain dehydrogenases/reductases (SDR)

(Jornvall et al., 1995). The situation is not so clear-cut with the HSDs, since they were shown

to belong to either the AKR or SDR superfamily. Until now, several of these pluripotent

HSDs have been described. The coincident ability of these pluripotent enzymes to metabolize

endogenous steroids and to reduce xenobiotic carbonyl compounds has evidently been in-

vented twice and subsequently been conserved during evolution in these two different protein

superfamilies, corresponding to the principle of convergent evolution (reviewed in (Maser,

1995)). At present, 12 different pluripotent carbonyl-reducing enzymes belonging to the

AKR and SDR protein superfamilies are known in humans (Matsunaga et al., 2006).

Sequence comparison, chemical modification and site-directed mutagenesis studies, combined

with crystallographic analyses and bioinformatic data, led to the identification of important

structural features and consensus sequences of these enzymes (see below).

1.1

The Aldo-Keto Reductase (AKR) Superfamily

The aldo-keto reductase (AKR) superfamily represents a growing superfamily of NADP(H)-

dependent oxidoreductases, which accept many structurally different substrates of endoge-

nous and exogenous origin. This protein superfamily includes, amongst others, aldose reduc-

tase (EC 1.1.1.21), aldehyde reductase (EC 1.1.1.2) and several HSDs (EC 1.1.1.x), some of

which have previously been designated dihydrodiol dehydrogenases (EC 1.3.1.20) (Jez et al.,

1997b). These enzymes are monomeric (

/)8-barrel proteins, about 320 amino acids in

length, which bind their cosubstrate without a Rossmann-fold motif (Jez et al., 1997b). The

active site of the AKR members contains a conserved tetrad of the amino acids Tyr, His, Asp

and Lys. Found in almost every living organism, the AKRs metabolize steroids, sugars, pros-

taglandins, polycyclic aromatic hydrocarbons, and a great variety of non-steroidal aldehydes

and ketones.

Introduction

The broad substrate specificity of some AKR proteins has led to some ambiguities, e.g. the

indomethacin-sensitive cytosolic rat liver 3

-HSD has previously also been described as di-

hydrodiol dehydrogenase, chlordecone reductase, or bile acid binding protein. To address

these issues, a new nomenclature system for the AKR superfamily has been established (Jez

et al., 1997b). Based on amino acid sequence identities of the different AKR members, the

superfamily currently comprises 14 families: AKR1 AKR14. Mammalian AKRs are found

predominantly in the AKR1 and AKR7 families, whereas members of the AKR2 AKR5 and

AKR8 AKR14 families have been identified in plants, yeast and bacteria. The AKR6 family

is formed by the -subunit of the Shaker-related voltage-gated K

-channel. An excellent

source of more information is available on the AKR superfamily homepage

(

www.med.upenn.edu/akr

) which has been created by TM Penning and coworkers. Many of

the mammalian AKRs are potential therapeutic targets, and structure-based drug design may

lead to compounds with the desired specificity and clinical efficacy.

1.2

Short-Chain Dehydrogenase/Reductase (SDR) Superfamily

The second superfamily to which carbonyl-reducing enzymes belong, the short-chain dehy-

drogenases/reductases (SDR), comprises a wide range of procaryotic and eukaryotic enzymes

involved in functions as diverse as the metabolism of steroids, sugars, aromatic hydrocarbons

and prostaglandins, the fixation of nitrogen and the synthesis of antibiotics (Persson et al.,

1991; Jornvall et al., 1995). The superfamily is made up of enzymes averaging 250 350

amino acids in length which function independently of metal cofactors, and its members are

readily distinguishable from those of the medium-chain and long-chain alcohol dehydro-

genase superfamilies (Persson et al., 1991; Jornvall et al., 1995). Among the members of the

various families and subfamilies of the SDRs, there are no structural similarities to the AKR

proteins in spite of overlapping substrate specificities (Flynn and Green, 1993; Maser, 1995).

They exist as soluble or, in some cases, membrane-bound proteins, mostly homodimers or

homotetramers (Ghosh et al., 1994b; Ghosh et al., 1995; Tanaka et al., 1996a; Tanaka et al.,

1996b; Benach et al., 1998). All SDR proteins share well-conserved primary structure ele-

ments that are restricted to certain segments in the sequence, indicating a possibly common

fold, active site, reaction mechanism, and cosubstrate and substrate binding regions (Persson

et al., 1991; Jornvall et al., 1995). Analysis of the three-dimensional structure of the SDRs

Introduction

characterized so far shows that both NAD(H) and NADP(H) bind to the classical motif of

the Rossmann fold which is characteristic of the coenzyme-binding domain of many other

dehydrogenases (Duax et al., 2000). It depends on their predominant function as a dehydro-

genase or reductase whether they use either NAD(P) or NAD(P)H as cosubstrate (cf. chap-

ter 2.3.3 of this review).

The SDRs comprise a large and highly divergent superfamily with about 3000 known forms,

including species variants (Persson et al., 2003). They share a residue identity of typically 15-

30% in pairwise comparison (Kallberg et al., 2002a), indicating early duplicatory origins and

extensive divergence (Jornvall et al., 1995). Nevertheless, the folding pattern of the members

whose structures have been solved so far is conserved with largely superimposable peptide

backbones (Krook et al., 1993b; Ghosh et al., 2001).

The SDRs have initially been subdivided into two large families, the "classical" with 250-odd

residues and the "extended" with 350-odd residues (Jörnvall et al., 1995). Based on patterns

of charged residues in the cosubstrate-binding region, these families have recently been clas-

sified in seven subfamilies of "classical" SDRs and three subfamilies of "extended" SDRs.

Three further families are novel entities, denoted "intermediate", "divergent" and "complex",

encompassing short-chain alcohol dehydrogenases, enoyl reductases and multifunctional en-

zymes, respectively (Persson et al., 2003).

One of the most important physiological functions of the SDR enzymes appears to be the

conversion of signalling molecules to either the active or inactive state (Krozowski, 1992).

II.3

Hydroxysteroid Dehydrogenases as Carbonyl Reductases

Hydroxysteroid dehydrogenases (HSDs) are pyridine nucleotide-dependent oxidoreductases

which mediate the interconversion of secondary alcohols and ketones. They play pivotal roles

in the biosynthesis and inactivation of steroid hormones. Steroid hormones act by binding to

receptor proteins in target cells, which leads to transcriptional regulation of different gene

products and the desired physiological response. In target tissues, HSDs convert potent steroid

hormones to their cognate inactive metabolites (and vice versa) and thus regulate the occu-

pancy of steroid hormone receptors (Penning et al., 1997; Duax et al., 2000). The reactions

Introduction

they catalyze are position- and stereospecific and usually involve the interconversion of a car-

bonyl and a hydroxyl group on either the steroid nucleus or side chain. In normal cell physi-

ology, HSDs therefore function as important prereceptor regulators of signalling pathways by

acting as "molecular switches" for receptor-active and receptor-inactive hormone (Monder

and White, 1993; Penning et al., 1996; Labrie et al., 1997). Because the enzymes have much

greater tissue specificity than the receptors, they have emerged as attractive targets for the

design of potent and selective drugs that combat steroid-related disorders. HSDs are found in

every organism investigated so far.

Interestingly, several HSDs exhibit other activities besides steroid oxidoreduction. In addition

to being specific for their physiological steroid substrate, they can catalyze the carbonyl re-

duction of a great variety of non-steroidal aldehydes, ketones and quinones (reviewed in

(Maser, 1995). Accordingly, HSDs participate in drug metabolism and play a significant role

in the defence of an organism against the deleterious effects of endogenous and exogenous

toxicants. Due to the fact that several enzymes of both superfamilies exhibit pluripotency for

steroidal and non-steroidal carbonyl substrates (Maser, 1995), AKR and SDR seem to be the

result of a convergent evolution.

Extensively characterized HSDs with carbonyl-reducing activity of the AKR superfamily are

-HSD (mammalian), 17-HSD (mammalian) and 20-HSD (mammalian and protozoan)

[reviewed in (Maser, 1995; Matsunaga et al., 2006)]. These HSDs are now classified within

the AKR1C-subfamily. Extensively characterized HSDs with carbonyl-reducing activity of

the SDR superfamily are 3

-HSD (bacterial), 3/20-HSD (mammalian and bacterial) and

11-HSD (mammalian) [reviewed in (Maser, 1995; Matsunaga et al., 2006)].

The sections below include recent findings on pluripotent HSDs/carbonyl reductases of the

SDR protein superfamily.

General Features of the SDR Superfamily Enzymes

III General Features of the SDR Superfamily Enzymes

III.1

Historical Background: Functional Characterization

Carbonyl reductases of the SDR superfamily accept some of the same substrates as do other

carbonyl-reducing enzymes from the AKR superfamily (Wermuth et al., 1982). SDR car-

bonyl reductases do also metabolize aromatic ketones and quinones (Wermuth, 1982). In gen-

eral, the metabolism of a broad spectrum of xenobiotic substrates results in their detoxifica-

tion or inactivation, since the chemically more reactive carbonyl compounds (aldehyde or

ketone) have been converted to less reactive and less lipophilic hydroxy intermediates that are

easier to conjugate and to eliminate (Felsted and Bachur, 1980; Wermuth, 1985; Wermuth et

al., 1986; Atalla and Maser, 1999).

Human carbonyl reductase (CBR1) had been supposed to be involved in prostaglandin and

steroid metabolism in some tissues (Lee and Levine, 1974a; Felsted and Bachur, 1980; Wer-

muth, 1981). Later, Wermuth (Wermuth et al., 1986) and Schieber (Schieber et al., 1992)

showed that, due to high non-physiological Km values and low turnover numbers of the en-

zyme, as well as due to low tissue concentrations of the prostaglandin or steroid substrates,

these compounds can at best be only poor endogenous substrates. In contrast, quinones from

polycyclic aromatic hydrocarbons were much better substrates than prostaglandins or steroids

(Wermuth et al., 1986). The best substrates for human CBR1 comprise K region ortho-

quinones of polycyclic aromatic hydrocarbons such as benzo[a]anthracene and

benzo[a]pyrene (Wermuth et al., 1986). Menadione and 9,10-phenanthrenequinone are excel-

lent carbonyl reductase substrates, and menadione is often used as a model quinone substrate

(Wermuth et al., 1986).

Previously, it was suggested that carbonyl reductases could be divided into two groups with

respect to their substrate specificity for endogenous prostaglandins and steroids (Ohara et al.,

1995). The first group, the carbonyl reductases of human tissues (Wermuth, 1981; Schieber et

al., 1992), pig kidney (Wermuth, 1981; Schieber et al., 1992), and rat ovary (Iwata et al.,

1989) have been demonstrated or suggested to be identical to prostaglandin 9-ketoreductase

or NADP

-dependent 15-hydroxyprostaglandin dehydrogenase. The second group comprised

hepatic carbonyl reductases of rats (Penning et al., 1984), rabbits (Sawada et al., 1980;

General Features of the SDR Superfamily Enzymes

Fischer et al., 1985), guinea pigs (Hara et al., 1986a) and mice (Sawada et al., 1988). This

group did also possess 3- or 17-hydroxysteroid dehydrogenase and dihydrodiol

dehydrogenase (DDH) (EC 1.3.1.20) activities and the individual enzymes have been thought

to be identical to hydroxysteroid dehydrogenases and/or DDHs (Ohara et al., 1995). As

pointed out above, these enzymes are today classified as members of the AKR superfamily

where they constitute the AKR1C subfamily (Hyndman et al., 2003; Penning, 2005; Matsu-

naga et al., 2006).

Lacking molecular tools, inhibitors have been used to characterize and to differentiate the

multiplicity of carbonyl-reducing enzymes. As inhibitors for carbonyl reductases, flavonoids

such as rutin, quercetin or quercitrin (Ahmed et al., 1979; Sawada and Hara, 1979; Sawada et

al., 1979; Felsted and Bachur, 1980) have been used. In contrast, phenobarbital served as a

diagnostic inhibitor of aldehyde reductase (Felsted and Bachur, 1980). Other inhibitors that

have been identified for two carbonyl reductases from rat ovary were indomethacin, fu-

rosemide and disulfiram (Iwata et al., 1989).

III.2

SDR Superfamily Classification

The superfamily of short-chain dehydrogenases/reductases (SDR) includes members of great

functional diversity, with sequence homologies of only 15-30%, but with very conserved spe-

cific sequence motifs as well as highly conserved folding patterns. The SDR superfamily was

established in 1981 (Jornvall et al., 1981). At that time, only two members were known to

belong to this family, prokaryotic ribitol dehydrogenase and an insect alcohol dehydrogenase.

Today, the SDR superfamily consists of more than 3000 members which cover a wide sub-

strate spectrum including alcohols, sugars, steroids, prostaglandins, aromatic compounds and

xenobiotics (Persson et al., 2003). Their general primary structure is composed of a cofactor-

binding region at the N-terminal part, a catalytic active site in the central part, a substrate-

binding region, and a C-terminal extension important for oligomerization. Despite the low

sequence identity, their three-dimensional structures superimpose very well, except for the C-

terminal region, where the diversity between the SDR proteins is much higher (Persson et al.,

2003).

General Features of the SDR Superfamily Enzymes

Initially, the SDR superfamily had been divided into 2 families, the "classical" and the "ex-

tended" SDRs. Classical SDRs consist of one-domain subunits of about 250 residues, whereas

members of the "extended" SDR family consist of one-domain subunits of about 350 resi-

dues. Members of both families catalyze NAD(P)(H)-dependent oxidation/reduction reac-

tions. Persson and coworkers (Persson et al., 2003) further characterized the two families. The

"extended" and "classical" families within the SDR superfamily differ, amongst others, in the

glycine-rich pattern of the cofactor-binding region (Jornvall et al., 1995). The three glycine

residues are characteristic for NAD(P)H-binding and are spaced differently in the two fami-

lies (Jornvall et al., 1981; Jornvall et al., 1999; Persson et al., 2003). Further differences occur

in the motif positioned close to strand 4, which is less conserved among "extended" SDRs

than among "classical" SDRs, as well as in the residues adjacent to the active site, a fact

which has been described in more detail in (Persson et al., 2003).

Persson and coworkers (Persson et al., 2003) used a trained hidden Markov model (Karplus et

al., 1998; Karplus et al., 2005) which was based on a multiple sequence alignment with

known SDRs as seed sequences to search the databases Swissprot (Apweiler et al., 2000; Bai-

roch and Apweiler, 2000) and KIND (Kallberg and Persson, 1999) for further members of the

SDR superfamily. They found that the SDR superfamily can be clustered into five instead of

two families, namely the "intermediate", "divergent" and "complex" families, in addition to

the already established "classical" and "extended" families. The overall features of these five

families of the SDR superfamily are summarized in (Table III-1). The carbonyl reductases

which we focus on in this article belong to the family of "classical" SDRs.

Carbonyl reductases which belong to the group of "classical" SDRs reduce aldehydes and

keto groups of prostaglandins, steroids, pterins, and quinones derived from polycyclic aro-

matic hydrocarbons. In addition, many keto drugs as well as the aliphatic keto side chain of

the anthracycline anticancer drugs daunorubicin and doxorubicin are reduced (Bachur, 1976;

Ahmed et al., 1978; Ahmed et al., 1981; Felsted and Bachur, 1982).

General Features of the SDR Superfamily Enzymes

family

feature

Divergent

Classical

Intermediate

Extended

Complex

Monomer size

~ 250 residues

~ 350 residues

Cofactors NADH

NAD(P)(H)

NAD(P)(H) NADP(H)

Catalytic reaction

Oxidation/reduction

Beta-ketoacyl

reduction

Cofactor binding

GxxxxxSxA (in

bacteria)

TGxxxGhG G/AxxGxxG/A

TGxxGhaG

Coenzyme bind-

ing

N-terminal N-terminal

N-terminal N-terminal N-terminal

Active site

YxxMxxxK

YxxxK

YxxxN

Substrate binding

C-terminal

C-terminal C-terminal

C-terminal

Sequence identity

(according to

swissprot)

28-99% 8-99%

27-99% 10-99% 20-74%

Examples

Enoyl reductases

from bacteria and

plants

Oxidoreductases (EC1.-.-.-),

steroid dehydrogenases

(including 11-HSD1, 11-

HSD2, 3-HSD/CR, 3/20-

HSD), carbonyl reductases

(including human CBR1 and

CBR3)

Drosophila

alcohol dehydro-

genase

Isomerases

(EC 5.-.-.-, galac-

tose, epimerases) ;

lyases (EC 4.-.-.- ,

glucose dehydrata-

ses)

oxidoreductases

Parts of multifunc-

tional enzyme

complexes (fatty

acid synthase)

Table III-1: Features of the five families within the SDR superfamily according to Persson and coworkers

(Kallberg et al., 2002b; Persson et al., 2003).

Each of the five SDR families contain conserved sequence patterns, where "a" signalises an aromatic residue,

"h" a hydrophobic residue, and "x" any residue. Amino acid residues are given in the one-letter code. Items with

no data mean that no information was available in the literature.

III.3

Structural Features of the SDR Members

As indicated above, the criterion for SDR membership is the occurrence of typical sequence

motifs which are arranged in a specific manner, called Rossmann fold (Rossmann et al.,

1975). These conserved residues play significant roles in the catalytically active site, the sub-

strate binding region, the cofactor binding motif, and the C-terminal extension (Fig. III-1 and

Fig. III-2).

General Features of the SDR Superfamily Enzymes

-b

-m

CR2

CR1

rtC

CR3

D1-

PHC

---

TKI

CCP

---

QFV

---

TKI

CCP

---

QFV

---

TKI

CCP

---

QFV

---

CCP

---

QFV

---

TKI

CCP

---

QFV

---

TKI

CCP

---

QFV

---

TKI

CCP

---

QFV

A-

---

SKL

KRK

CCP

I-

-Q

---

SKL

KRK

CCP

---

-Q

8 L

AAK

SSI

DGG

KSL

QRI

-G

---

S-

-K

YDG

T-

AW-

---

QSN

SSL

A-

AAP

LWT

TSY

NK-

---

QSN

SSM

T-

AAP

SWT

RSY

N-

---

QSN

SSL

T-

ASP

TPI

N-

---

SSV

T-

ASP

AEY

A 300

7 E

R-

---

GG-

SSV

P-

---

VGG

G-

-T

ASR

L-

1 D

R-

---

GVP

T-

---

RSA

VDA

G-

AS-

1 S

R-

---

GVP

T-

---

RSA

VDA

G-

AS-

4 A

---

AGG

G-

LAL

TGI

---

GNY

TSS

VDG

G-

-W

GPT

Q 2

0 A

G-

---

-H

SSV

-GG

ATE

GLQ

G-

---

IAK

IDG

RPT

VRA

SCSPE

LRA

NCSPE

LRA

NCSLE

LWA

NCSPE

LRA

SCSPE

RRA

SCSPE

TEGVRA

LNECSPE

QCLRA

NCSED

TEG

QGSKA

NCSED

CNT

TEG

HQK

HEK

HAK

HRK

HER

DTK

PSSA

PNSA

SDST

PNSP

PDSA

ral

op-

aloo

ata

tic

tria

lys

odi

NAD

-spe

cif

V-

DFL

G-

-G

VTM

G-

-A

DFL

G-

-G

VTM

G-

-A

DFL

G-

-G

VTM

G-

DFL

R-

-G

VTM

D-

DFL

G-

-G

VTM

G-

DFL

G-

-G

VTM

G-

V-

DFL

G-

-G

LTM

V-

DFL

G-

-G

MTL

A-

-E

DFL

G-

-G

MTL

A-

-M

NRE

P 12

M-

AQA

G-

-G

N-

KSM

-M

G-

M-

VEA

G-

-G

S-

RSM

M-

VKA

G-

-G

S-

M-

EAL

AEA

G-

-G

T-

KSM

D-

AMA

G-

-G

P 13

-M

SGL

S-

--T

S-

--A

P-

-M

SGL

S-

--T

S-

--A

P-

-N

A-

E--

G-

-S

F-

P 12

-M

A-

VIA

-Q

---

-G

G-

-P

LGN

P 99

som

rget

ing

gnal

CBR1

CHC

rtC

CBR3

CHC

D1-

D1-r

D1-m

D1-

General Features of the SDR Superfamily Enzymes

Fig. III-1: Multiple sequence alignments of some well-known members of the SDR superfamily

Abbreviations: CBR1, CBR3 (human carbonyl reductase 1 and 3); CHCR1, CHCR2, CHCR3 (Chinese hamster

carbonyl reductases 1-3); iCR, nCR, rtCR (inducible, non-inducible and rat-testis carbonyl reductase); Sniffer

(carbonyl reductase Sniffer from Drosophila); 11-HSD1-h, 11-HSD1-m, 11-HSD1-r, 11-HSD1-g (11-

HSDs from human, mouse, rat and guinea pig); PHCR (pig heart carbonyl reductase); MLCR (mouse lung car-

bonyl reductase); PLCR (pig lung carbonyl reductase); 3/20-HSD (S. hydrogenans); 3-HSD/CR (C. testos-

teroni).

The alignment indicates the existence of an extraloop-domain upstream of the conserved Tyr-x-x-x-Lys motif

within the monomeric enzymes from human, Chinese hamster and rat. This extraloop-domain can also be found

in the dimeric 3-HSD/CR from C. testosteroni. Monomeric enzymes are marked in blue, dimeric enzymes in

red and tetrameric enzymes in yellow. Colored residues represent conservation of 50% identity. Alignment was

performed using the clustalW- and bioedit-program.

The overall structure of the SDR superfamily members is built up of a sequence of alternating

-helices and -strands. These -strands form a four- or five-stranded parallel -sheet with

two or three -helices residing on either side. This dinucleotide binding motif composed of

units is called Rossmann fold (Rossmann et al., 1975). The Rossmann fold in the SDR

superfamily is very stable, tolerating, despite little sequence conservations, some mutations at

every site of the folding motif without loss of function (Duax et al., 2000). It is less readily

understood how a protein conformation as consistent as the Rossmann fold has no structurally

conserved residues. Once substrate specificity evolved, changes in the active site could lead to

loss of function or change in specificity. It appears that in the SDR superfamily, function

evolved early, while specificity evolved later (Duax et al., 2000).

III.3.1

Catalytic Triade and Catalytic Mechanism

Already in 1981, the importance of the Tyr-x-x-x-Lys segment as the most conserved se-

quence motif of the SDRs became apparent (Jornvall et al., 1981). Nearly all "classical" SDR

enzymes use the conserved Tyr-Lys-Ser motif as catalytic residues (Jornvall et al., 1995),

whereas the enzymes of the "divergent" use the sequence motif Tyr-x-x-Met-x-x-x-Lys with

the tyrosine being positioned three amino acids upstream as compared to Tyr-x-x-x-Lys. The

subfamily of the "complex" SDRs have the unique motif Tyr-x-x-x-n (where n means any

other amino acid) at the active site (Persson et al., 2003). The amino acids which constitute

the catalytic triad are critically important for enzyme function. They appear to maintain a

fixed position relative to the scaffolding of the -folding and the cofactor position.

General Features of the SDR Superfamily Enzymes

Catalytic triade

Cofactor binding

Extraloop

Helices of the four helix bundle (Q-interface)

Fig. III-2: Folding topology of 3-HSD/CR from C. testosteroni.

The conserved structural elements involved in oligomerization are named only for the upper one of the two sub-

units. Note that the four helix bundle (Q-interface; blue) is blocked by the extraloop-domain (yellow). Therefore,

oligomerization (dimerization) takes place only via helix G and strand G (P-interface; purple).

Most likely, all SDRs share a common reaction mechanism which follows a compulsory or-

dered pathway with the cofactor binding first (Grimm et al., 2000b). After the hydroxyl or

carbonyl substrate binds, the so-called substrate binding loop becomes well ordered and cov-

ers the substrate as well as the catalytic center from the aqueous environment. This substrate

binding loop is highly flexible in most apo-structures of SDRs. Ghosh and coworkers (Ghosh

et al., 1994b) proposed a model for 3/20-HSD in which the conserved residues Tyr, Lys

and Ser, together with solvent molecules, could catalyze the reaction in the catalytic cavity

(Fig. III-3). The Tyr hydroxyl has been proposed to be the proton donor in an electrophilic

attack on the substrate carbonyl in a reduction reaction (Ghosh et al., 1994b). The positively

charged side chain of Lys is positioned in proximity to the hydroxyl oxygen of Tyr and facili-

tates the proton transfer. The Ser also participates in catalysis by stabilizing the reaction in-

termediates or as part of a proton-relay network. After the proton is transferred from Tyr to

General Features of the SDR Superfamily Enzymes

the 20-keto oxygen of the steroid, it can be replenished by the solvent surrounding the resi-

dues in the catalytic cavity (Ghosh et al., 1994b; Duax et al., 2000). The substrate is oriented

by the catalytic Ser which stabilizes the transient reaction intermediate (Grimm et al., 2000b).

The conserved Lys-residue plays a dual role in the catalytic triad. It orients the cofactor by

forming hydrogen bonds to the nicotinamide-ribose moiety and lowers the pK

of the catalytic

Tyr via electrostatic interaction (Ghosh et al., 1994b; Auerbach et al., 1997; Benach et al.,

1999).

Fig. III-3: Reaction mechanism and catalytic triade exemplified by 3/20-HSD from S. hydrogenans

S139

C H

Y152

K156

C ONH

General Features of the SDR Superfamily Enzymes

III.3.2

Substrate Binding and Substrate-Binding Loop

Most of the amino acid residues which are partially conserved within the SDR superfamily

members are located in the core of the Rossmann fold and are primarily hydrophobic (Duax et

al., 2000). The only conserved residues in the substrate-binding cleft are those of the catalytic

triad: Ser, Tyr and Lys. No conserved residues can be found in the substrate-binding site. This

is indicative for the fact that, for example, the HSDs show high selectivity for their specific

physiological steroid substrate. However, for the carbonyl reductases, specific endogenous

substrates have not been identified until now, they rather display broad substrate specificities,

ranging from steroids, prostaglandins, sugars and alcohols to xenobiotic aldehydes and ke-

tones (Duax et al., 2000).

Usually, SDR enzymes have a substrate-binding loop of more than 20 residues which covers

the active site and becomes well ordered after the hydroxyl or carbonyl substrate binding.

This loop region is located between -strand F (F) and -helix G (G) and is very often dis-

ordered in apo-structures (without substrate) of SDR enzymes. After substrate binding, the

loop shows a well-defined, mainly -helical, geometry. In Sniffer from Drosophila no -

helical secondary structure element is apparent in the substrate-binding loop (Sgraja et al.,

2004). Here, in the presence of NADP

, only a short loop could be detected in the electron

density which connects strand F and helix G (Sgraja et al., 2004) (Fig. III-4) .

When the -carbon backbone of known SDR structures is superimposed, the conserved resi-

dues appear at the core of the structure and in the cofactor-binding domain, but not in the sub-

strate-binding pocket (Duax et al., 2000). Contrary to the variability in the substrate-binding

pocket, the association of the Rossmann fold with the cofactor binding is very consistent and

the NADP(H) binding domains exhibit architectural integrity, despite having sequence varia-

tions (Duax et al., 2000). The variability in the substrate-binding pocket is plausible when

keeping in mind the variety of structures used as substrates by different members of the SDR

superfamily (Duax et al., 2000).

General Features of the SDR Superfamily Enzymes

III.3.3

Cofactor Binding

In the N-terminal part of the SDR proteins, a glycine-rich pattern important for cofactor bind-

ing has been identified (Jornvall et al., 1995). This conserved sequence motif includes three

glycine residues which are located at comparable points in their sequences and form a turn

between a -strand and an -helix that border the cofactor-binding site (Duax et al., 2000).

Structural analysis has shown that the glycine residues interact with the pyrophosphate moiety

of the cofactor. The individual subfamilies of the SDR superfamily differ regarding the spac-

ing of the three glycine residues, where the Gly-x-x-x-Gly-x-Gly motif, for example, is char-

acteristic for the subfamily of the "classic" SDRs (Jornvall et al., 1981; Jornvall et al., 1999).

The use of either reduced or oxidised NAD(P) as cofactor depends on their predominant func-

tion as a dehydrogenase or reductase of individual members of the SDR superfamily. In addi-

tion, both NAD(H) and NADP(H) bind to the classical motif of the Rossmann fold. The

specificity for NADP(H), rather than to NAD(H), is conferred by two highly conserved basic

residues in the N-terminal part of the peptide chain. Tanaka et al. (Tanaka et al., 1996a) iden-

tified two basic residues (Lys-17 and Arg-39) in the tetrameric mouse lung carbonyl reductase

(MLCR) which promote NADPH binding by interaction with the 2'-phosphate group. These

residues are believed to confer specificity for NADPH in other SDRs as well.

Human monomeric carbonyl reductase (CBR1) exhibits less than 25% overall sequence iden-

tity with tetrameric MLCR (Jornvall et al., 1995). However, similar to MLCR, human CBR1

has distal basic (Lys-15 and Arg-38) and neutral (Ala-37) residues at positions corresponding

to Lys-17, Arg-39 and Thr-38 in tetrameric MLCR (Tanaka et al., 1996a) and is strictly

NADPH-dependent (Wermuth et al., 1988; Sciotti and Wermuth, 2001). Sciotti and Wermuth

(Sciotti and Wermuth, 2001) showed that a positive charge at positions 15 and 38 and the ab-

sence of a negative charge at position 37 significantly contribute to NADPH specificity. SDR

enzymes which are NAD-dependent lack these two basic residues. Instead, an Asp which is

adjacent to the position of the distal basic residue in NADPH-dependent enzymes appears to

determine NAD specificity (Sciotti et al., 2000). Sequence alignments of NAD-dependent

enzymes with the NADPH-dependent tetrameric and monomeric carbonyl reductases show

that Ala-37 of the monomeric and Thr-38 of the tetrameric carbonyl reductase correspond to

the Asp residue which plays a role in NAD specificity. Sciotti and Wermuth (Sciotti and

Wermuth, 2001) substituted Asp for Ala-37 in the strictly NADPH-dependent human CBR1.

Unexpectedly, the Asp residue had little effect on the NADH-dependent activity and NADPH

General Features of the SDR Superfamily Enzymes

was still the preferred substrate of the mutant Ala37Asp. Therefore, factors other than only

the absence of Asp at position 38 must be responsible for the low efficiency of monomeric

human CBR1 in the presence of NADH Fig. III-1.

III.3.4

C-terminal Extension and 3

-Helices

Many enzymes of the SDR superfamily contain a C-terminal extension consisting of several

-helices. Examples are human estrogenic 17-HSD (Ghosh et al., 1995) and the Drosophila

enzymes alcohol dehydrogenase (DADH) and Sniffer. Benach and coworkers showed in

DADH that these C-terminal helices, which are of more variable character, are main elements

of dimer association and that these structural features have additional importance in closing of

the active site cavity upon cofactor and substrate binding (Benach et al., 1998).

-helices are structural elements which are present as -helical extensions in loops and as

connectors between -strands (Pal and Basu, 1999). In many cases they occur independently

of any other neighbouring secondary structural elements and form novel super-secondary

structural motifs. These motifs have possible implications for protein folding, local conforma-

tional relaxations and biological functions (Pal and Basu, 1999).

The Sniffer protein has both structural features which contribute together to dimer formation.

At the C-terminal end of the Sniffer protein there is a short loop with a sharp turn directly fol-

lowing strand G. This loop runs perpendicular to the central -sheet of the Rossmann fold

and touches the 3

-helix following strand E. The C-terminal ends of the two monomers ap-

proach each other and the indole moiety of the Trp-248 interacts through its planar ring sys-

tem with the side-chain of Ile-159 which is rooted at the 3

-helix of the second subunit

(Sgraja et al., 2004).

III.3.5

Oligomerization and Interfaces

The enzymes of the SDR superfamily can be divided into different groups concerning their

oligomerization behaviour, their subcellular localization and the existence of structural ele-

ments with significant effects on protein folding. Most enzymes of the superfamily appear as

either homodimers, e.g. Sniffer (Sgraja et al., 2004), DADH (Benach et al., 1998) and dihy-

General Features of the SDR Superfamily Enzymes

dropteridine reductase (DHPR) (Varughese et al., 1992), or homotetramers like 3/20-HSD

(Ghosh et al., 1994b) and mouse lung carbonyl reductase (Tanaka et al., 1996a). Only few

members occur as monomers like for example human carbonyl reductase (CBR1) (Wermuth,

1981) and pig testicular carbonyl reductase (PTCR) (Ghosh et al., 2001). In spite of the low

residue identities between the SDR members, the folding pattern of the individual (sub-) units

is highly conserved with largely superimposable peptide backbones (Krook et al., 1993b;

Ghosh et al., 2001) (Fig. III-4).

Two types of oligomerization, involving different structural elements, can be found. One type

of monomer association is the so-called P-axis interface comprising the antiparallel associa-

tion of -strand G and -helix G of each subunit (Ghosh et al., 1994b). This type of oli-

gomerization is usually found in homotetrameric SDRs and has only been observed in one

dimeric SDR, namely 3-HSD/CR from C. testosteroni (Grimm et al., 2000a). The P-axis

interface does not involve significant hydrophobic interactions (Ghosh et al., 1994b) and is

less critical to the structural integrity of the active site than the second type of oligomeriza-

tion, the so-called Q-axis interface.

The Q-axis interface consists of -helix E and -helix F of each of the two subunits to form a

bundle of four helices. This four helix bundle represents the most important structural motif

of the Q-axis interface, especially in dimeric SDR proteins. It is predicted to be important for

the integrity of the active site clefts and has been shown to be critical to function (Puranen et

al., 1997; Ghosh et al., 2001). In addition, although more than half of the crystal structures of

SDRs exhibit the same quaternary structure, only the four helix bundle as the main structural

feature of the Q-axis interface that involves the largest surface area of association (Ghosh et

al., 1994b) is conserved among the SDRs.

General Features of the SDR Superfamily Enzymes

SBL

extraloop

EF1

EF2

EF3

Fig. III-4: Typical Rossmann fold of SDR enzymes exemplified by 3-HSD/CR from C. testosteroni.

3-HSD/CR contains an unordered substrate binding loop (SBL) and an extraloop-domain between strand E

and helix F which normally are part of the four helix bundle as the main structural feature of the oligomeriza-

tion Q-interface.

The dimerization interface is close to the active site and stabilizes the interior of the molecule.

Tsigelny and Baker postulated that the neighbouring amino acids of the conserved tyrosine

and lysine residues in the active site of SDR family members might interact with the hydro-

phobic outer surface of the F helix and hence stabilize the dimer interface of the proteins

(Tsigelny and Baker, 1995b). Therefore, it is suggested that monomerization of dimeric or

tetrameric proteins in the SDR family most likely abolishes their enzyme activity. Hoffmann

and coworkers (Hoffmann et al., 2006) redesigned the region of the interfacial four helix bun-

dle in the dimeric 3-HSD/CR from C. testosteroni, which comprises an additional -helical

subdomain preventing the enzyme of dimerization along the typical Q-axis interface. Their

data of a soluble and active protein will give further insights into the requirements of a protein

for the architecture of the active site and the structural elements comprising the Q-axis inter-

face.

In the dimeric human 17-HSD, Puranen and coworkers (Puranen et al., 1997) substituted

amino acids of the four helix bundle within the hydrophobic Q-axis interface resulting in inac-

tive multi-aggregates of the protein, while neither dimers nor monomers were detectable.

Ghosh and coworkers were also able to show that the four helix bundle is critical for protein

General Features of the SDR Superfamily Enzymes

function. Disruptions of the helix arrangement had negative effects on the integrity of the

catalytic centres of the oligomerizing monomers in 3/20-HSD of S. hydrogenans (Ghosh et

al., 2001).

An example for a cytosolic and monomeric carbonyl reductase is human CBR1. The reason

for the monomeric structure of this enzyme is a so-called additional extraloop-domain be-

tween the Q-interfacial strand E and helix F which is believed to prevent oligomerization

along the Q-axis interface built up of the four helix bundle. Other monomeric carbonyl reduc-

tases include this -helical subdomain as well, but with different length. Among the mono-

meric carbonyl reductases with an additional subdomain are pig testicular carbonyl reductase

(PTCR) which has a sequence identity to human CBR1 of nearly 85%, as well as the cytosolic

and monomeric carbonyl reductases of rat (non-inducible, nCR; inducible, iCR; testis, rtCR)

and Chinese hamster (CHCR).

The group of tetrameric carbonyl reductases is not as consistent as the group of monomeric

carbonyl reductases. Tetrameric carbonyl reductases can be found on the subcellular level in

the mitochondria (namely carbonyl reductases from mouse and pig lung) (Nakayama et al.,

1986; Oritani et al., 1992) and in the cytoplasm, i.e. 3/20-HSD from S. hydrogenans

(Ghosh et al., 1992; Ghosh et al., 1994a; Ghosh et al., 1994b). The tetrameric carbonyl reduc-

tase from pig liver is a peroxisomal protein which contains the SRL-tripeptide (serine-

arginine-leucine) as a peroxisomal targeting signal (Usami et al., 2003) (cf. chapter IV ).

The Q-interfacial dimer interface of which Sniffer provides a typical example is also present

in tetramers (3/20-HSD). This interface is stabilized by interactive associations of symme-

try related -helices E and F. It has been shown that while conservative variations in amino

acids (i.e. substitution of one hydrophobic residue for another) is tolerated, non-conservative

mutations can disrupt dimer formation and cause loss of function (Duax et al., 2000).

Pluripotent Carbonyl Reductases of the SDR Superfamily

A special case represents dimeric 3-HSD/CR from C. testosteroni. This enzyme does not

undergo oligomerization along the Q-axis interface, because the protein contains an -helical

extraloop-domain within this Q-interfacial four helix bundle. As with the monomeric carbonyl

reductases containing this -helical extraloop-domain, oligomerization along this interface

region is prevented because of the disruption of the Q-axis interface. Most interestingly, 3-

HSD/CR performs dimerization along the P-axis interface (Grimm et al., 2000b). However,

this type of oligomerization via the P-axis interface can only be found in tetrameric SDRs

such as for example in 3/20-HSD from S. hydrogenans (Ghosh et al., 1994b) (see above).

IV Pluripotent Carbonyl Reductases of the SDR Superfamily

Carbonyl-reducing enzymes can be found ubiquitously in nature and they have been shown to

catalyze the NADPH-dependent reduction of a wide variety of biologically and pharmaco-

logically active substrates, as well as a variety of endogenous and xenobiotic carbonyl com-

pounds including quinones (Felsted and Bachur, 1980; Wermuth, 1985; Ohara et al., 1995;

Oppermann and Maser, 2000; Breyer-Pfaff et al., 2004; Breyer-Pfaff and Nill, 2004; Maser,

2004; Maser et al., 2005; Hoffmann et al., 2006; Martin et al., 2006; Maser et al., 2006). As

suggested for two-electron transferring quinone reductases (Lind et al., 1982), carbonyl re-

ductase might function as a cellular mechanism to control the toxicity of quinones, especially

in humans (Wermuth et al., 1986). In addition, reactive aldehydes derived from lipid peroxi-

dation were found to be substrates for the enzyme (Doorn et al., 2004; Hoffmann et al.,

2006). Interestingly, activity toward endogenous prostaglandins and steroids has been demon-

strated, although the low catalytic efficiency reported raised skepticism that these compounds

are physiological substrates of carbonyl reductases and that the corresponding pathways rep-

resent a physiological role for the enzymes (Chang and Tai, 1981; Wermuth, 1981; Jarabak et

al., 1983; Forrest and Gonzalez, 2000). Therefore, a general detoxification role for foreign

and metabolically derived carbonyl compounds seems to account for the enzymes' wide sub-

strate specificity and its ubiquitous distribution in human tissues (Wirth and Wermuth, 1992).

An evolutionary classification of the SDR superfamily members involved in carbonyl reduc-

tion is shown in Fig. IV-1.

Pluripotent Carbonyl Reductases of the SDR Superfamily

11 HSD2-g

11 HSD2-h

11 HSD2-r

11 HSD2-m

Sniffer

PTCR

CBR1

CHCR2

CHCR1

nCR

rtCR

iCR

CBR3

CHCR3

PHCR

MLCR

PLCR

3/20-HSD

3-HSD

11 HSD1-g

11HSD1-h

11HSD1-r

11 HSD1-m

osom

monomer

NADPH-dependent

lic

dimer

rosom

tet

NADH-dependent

lic

mer

Fig. IV-1: Evolutionary tree of carbonyl-reducing enzymes from different species

The nomenclature is according to Fig.2.

Pluripotent Carbonyl Reductases of the SDR Superfamily

IV.1

Carbonyl Reductases in Non-Mammals

Carbonyl reduction in microorganisms was mainly investigated by chemists who are inter-

ested in the production of optically active alcohols (Roberts, 1997; Faber, 1998). However,

the studies were generally performed with crude cell extracts and the specific enzyme in-

volved was often unclear. Only a few papers reported serious work on the purification, identi-

fication and characterization of carbonyl reductases and aldehyde reductases from microor-

ganisms such as Saccharomyces cerevisiae (Ypr1p, 2-methylbutyraldehyde reductase) (Ford

and Ellis, 2001; Ford and Ellis, 2002), Sporobolomyces salmicolor, Candida magnoliae

(NADPH-dependent aldehyde reductase, ARII) (Kita et al., 1999), Geotrichum candidum

(Matsuda et al., 2000), Rhodococcus erythropolis (carbonyl reductase, RECR) (Zelinski et al.,

1994), and Escherichia coli (2,5-diketo-D-gluconate reductase) (Oh et al., 1999; Yum et al.,

1999a; Yum et al., 1999b; Habrych et al., 2002; Thibault et al., 2004). In the future, reduction

of carbonyl compounds by microorganisms might gain great importance for biotechnological

uses (Ellis, 2002) or in the field of environmental sciences.

IV.1.1

/20-Hydroxysteroid Dehydrogenase of Streptomyces hydrogenans

Streptomyces hydrogenans belongs to the actinomycetes which are common in soil, plant de-

bris, dung, house dust, and many other habitats (Waksman, 1961; Backlund et al., 2005).

3/20-Hydroxysteroid dehydrogenase (3/20-HSD) from S. hydrogenans is an NADH-

linked enzyme which is involved in the reversible oxidation of the 3-group of androstane

derivatives and the 20-group of pregnane derivatives (Sweet and Samant, 1980). Interest-

ingly, although there is no known steroid hormone function in bacteria, the S. hydrogenans

gene encodes a hydroxysteroid dehydrogenase enzyme (Duax et al., 2000).

The structure of 3/20-HSD was the first structure solved and determined by X-ray diffrac-

tion of an enzyme belonging to the SDR superfamily and the third structure of an enzyme for

which steroids are the substrate (Ghosh et al., 1991). The active form of the enzyme is a

tetramer. The four identical monomers each consist of 253 amino acids with a calculated mo-

lecular mass of around 26 kDa and containing a single domain incorporating the active site

and the Rossmann fold cofactor binding motif (Ghosh et al., 1994b).

In a proposed catalytic mechanism of the 3/20-HSD enzyme, the Tyr-152 hydroxyl proton

initiates the electrophilic attack on the 20-keto oxygen of the steroid (Duax et al., 2000). The

Pluripotent Carbonyl Reductases of the SDR Superfamily

positively charged side chain of Lys-156 lies in close proximity to the Tyr-152 hydroxyl oxy-

gen and facilitates proton transfer. Ser-139 participates in catalysis by stabilizing the reaction

intermediates and as part of a proton-relay network, in which a proton, once transferred from

Tyr-152 to the 20-keto oxygen of the steroid, can be replenished by the solvent network sur-

rounding these residues in the catalytic cavity (Ghosh et al., 1994b; Duax et al., 2000). This

mechanism of catalysis does also apply to other members of the SDR superfamily.

IV.1.2

-Hydroxysteroid Dehydrogenase/Carbonyl Reductase of Comamonas testos-

teroni

Procaryotic 3

-hydroxysteroid dehydrogenase/carbonyl reductase (3-HSD/CR) has been

first described in the gram-negative bacterium Pseudomonas testosteroni which was later re-

named Comamonas testosteroni and reclassified as belonging to the

2-group of the Proteo-

bacteria (Michel-Briand, 1969; Michel-Briand and Roux, 1969; Skalhegg, 1975; Aukrust et

al., 1976; Shikita and Talalay, 1979; Tamaoka et al., 1987; Oppermann et al., 1993; Suzuki et

al., 1993; Abalain et al., 1995; Floch et al., 1995). C. testosteroni has been first isolated from

soil, mud and water, but has also been found in clinical isolates of patients suffering from

purulent meningitis (Arda et al., 2003), endocarditis (Heredia et al., 2004) and cystic fibrosis

(Coenye et al., 2002).

These bacteria are strictly aerobic, nonfermentative and chemoorganotrophic. They rarely

attack sugars, but grow well on organic acids and amino acids (Willems et al., 1992). In addi-

tion, C. testosteroni is able to grow on steroids as sole carbon source (Talalay et al., 1952).

3-HSD/CR is one of the first enzymes of the steroid catabolic pathway and, therefore, plays

a central role in steroid metabolism in C. testosteroni. The enzyme mediates the oxidoreduc-

tion at position 3 of the steroid nucleus of a great variety of C

19-27

steroids. This reaction is of

importance in the initiation of the complete degradation of these relatively inert substrates and

may significantly contribute to the bioremediation of hormonally active compounds in the

environment. The enzyme is also involved in the degradation of fusidic acid, thereby provid-

ing resistance of C. testosteroni towards this steroid antibiotic produced by the fungus

Fusidium coccineum (Oppermann et al., 1996).

Pluripotent Carbonyl Reductases of the SDR Superfamily

In addition to steroids, 3-HSD/CR accepts as substrates a wide spectrum of non-steroid

xenobiotic carbonyl compounds such as the cytochrome P450 inhibitor metyrapone, a

metyrapone-based class of environmentally safe anti-insect agents, e.g. NKI 42255 (2-(1-

imidazolyl)-1-(4-methoxyphenyl)-2-methyl-1-propanone) (Darvas et al., 1991; Belai et al.,

1995; Oppermann et al., 1996) (cf. chapter V.8 ; Detoxification of Insecticides), or the sig-

nificantly smaller p-nitrobenzaldehyde (Oppermann and Maser, 1996; Mobus and Maser,

1998).

Due to this fact, the enzyme's name was extended from 3

-hydroyxsteroid dehydro-

genase (3

-HSD) to 3-hydroyxsteroid dehydrogenase/carbonyl reductase (3-HSD/CR).

Interestingly, the expression of 3-HSD/CR in C. testosteroni is inducible by steroids (Möbus

et al., 1997). Steroid-induced bacterial cells exhibit a significantly faster degradation of ster-

oids, as well as a faster uptake and metabolism of NKI 42255 (Oppermann et al., 1996).

Therefore, induction by steroids results not only in an increase of resistance towards the ster-

oid antibiotic fusidic acid, but also in an enhancement of insecticide detoxification

(Oppermann et al., 1996).

3-HSD/CR uses NAD(H) as cofactor, depending on the reaction direction. Biochemical and

structural analysis revealed that the enzyme appears as a homodimer with a mode of oli-

gomerization along the P-axis interface analogous to tetramerization of 3/20-HSD from S.

hydrogenans (Ghosh et al., 1994b). Until then, this way of oligomerization had never been

observed in a dimeric SDR enzyme, where dimerization usually takes place via the Q-axis

interface (Grimm et al., 2000a), e.g. in Drosophila melanogaster alcohol dehydrogenase

(Benach et al., 1998) and Sniffer (Sgraja et al., 2004) (cf. chapter III.3.5; Oligomerization and

Interfaces).

IV.1.3

Insect Carbonyl Reductase: Sniffer of Drosophila melanogaster

The Drosophila melanogaster carbonyl reductase Sniffer is a NADPH-dependent member of

the SDR superfamily. The enzyme shows similarity with human carbonyl reductases CBR1

(27,4%) and CBR3 (24,9%) (Botella et al., 2004).

Interestingly, Sniffer knock-out mutants had a reduced lifespan, showed motoric dysfunctions

and exhibited age-related and oxidative stress-induced neurodegeneration (Botella et al.,

Pluripotent Carbonyl Reductases of the SDR Superfamily

2004). All symptoms of the altered phenotype resembled those in patients with neurodegen-

erative dysfunction. Histological analysis of the central nervous system showed an increased

vacuolization and apoptosis in the brain of mutant flies compared to wild-type flies. In con-

trast, overexpression of the carbonyl reductase Sniffer not only rescued the phenotype but

even induced a prolonged survival of neurons under hyperoxia conditions in wild-type flies

(Botella et al., 2004). These results suggested an important role of the carbonyl reductase

Sniffer in the neuroprotection against oxidative stress.

Besides being structurally related to CBR1, the Sniffer protein exhibited carbonyl reductase

activities with the CBR1 substrates 9,10-phenanthrenequinone, p-nitrobenzaldehyde, pyri-

dine-4-carboxaldehyde and menadione (Botella et al., 2004). Since human CBR1, in turn, was

shown to catalyze the carbonyl reduction of 4-oxononenal (Doorn et al., 2004), it is antici-

pated that the mechanism of neuroprotection by Sniffer in Drosophila is due to the detoxifica-

tion of reactive aldehydes derived from lipid peroxidation after oxidative stress (Hoffmann et

al., 2006) (cf. chapter V.3 ; Neuroprotection by Carbonyl Reductase?).

Gel-filtration experiments and the crystal structure revealed unambiguously that Sniffer ap-

pears as a dimer in solution, with a molecular mass of each monomer of 28 kDa. In the Sniffer

protein the two long helices E and F from two adjacent protein molecules associate to form

the four helix bundle (Sgraja et al., 2004) (cf. chapter III.3.5; Oligomerization and Interfaces).

This mode of oligomerization is observed in most other dimeric SDRs whose structures have

been solved so far (Ghosh et al., 1994b; Tanaka et al., 1996a; Benach et al., 1998). The only

exception of this rule is 3-HSD/CR from C. testosteroni, where dimerization occurs via an

interface region observed only in homotetrameric SDRs (Grimm et al., 2000a) (cf. chapter

III.3.5; Oligomerization and Interfaces).

IV.2

Carbonyl Reductases in Mammals

IV.2.1

Monomeric Cytosolic NADPH-Dependent Carbonyl Reductases

IV.2.1.1 Human Carbonyl Reductase 1 (CBR1)

Pluripotent Carbonyl Reductases of the SDR Superfamily

Human carbonyl reductase (CBR1) is a cytosolic and monomeric member of the SDR super-

family (Wermuth et al., 1988; Wirth and Wermuth, 1992; Forrest and Gonzalez, 2000). The

enzyme is widely distributed in human tissues, such as liver, epidermis, stomach, small intes-

tine, kidney, neuronal and glial cells of the CNS, and smooth muscle fibres (Wirth and Wer-

muth, 1985; Wirth and Wermuth, 1992; Forrest and Gonzalez, 2000). Small amounts of

CBR1 were found e.g. in cerebellum, oral cavity, oesophagus, kidney and skeletal muscle

(Wirth and Wermuth, 1992)

One of the first isolations of this enzyme was from human brain (Ris and von Wartburg,

1973). According to its broad substrate specificity, CBR1 from human brain was previously

also designated as secondary alcohol-NADP

oxidoreductase (EC 1.1.1.184) (Wirth and

Wermuth, 1992) and aldehyde reductase 1 (Ris and von Wartburg, 1973). In 1981, Wermuth

suggested the name carbonyl reductase (CBR1) to better describe the reduction of a wide va-

riety of carbonyl substrates (Wermuth, 1981; Wermuth et al., 1986). The substrate specificity

of human CBR1 suggested initially that the enzyme is an aldo-keto reductase. However,

structural investigations revealed no significant homologies to the aldo-keto reductases but, in

contrast, indicated the relationship to the short-chain dehydrogenase (SDR) superfamily

(Bohren et al., 1989).

Substrate diversity

Wermuth and coworkers found that the enzyme reduces a number of biologically and phar-

macologically active carbonyl compounds. During catalysis the pro 4S hydrogen atom of the

nicotinamide ring of NADPH is transferred to the substrate. Human CBR1 exhibits relatively

poor reductive activity towards aliphatic and alicyclic ketones, with the exception of daun-

orubicin and the glutathione adduct of prostaglandin A

(Wermuth, 1981).

The best substrates are quinones, e.g. menadione, ubiquinone-1, phenanthrenequinone and

tocopherolquinone followed by ketoaldehydes (e.g. phenylglyoxal), aromatic aldehydes con-

taining an electron-withdrawing substituent, e.g. 4-nitrobenzaldehyde or methylglyoxal, and

the biogenic aldehydes, indol-3-acetaldehyde and 4-hydroxyphenylacetaldehyde (Wermuth,

1981). Quinones, some of which are substrates of human CBR1, play important roles as oxi-

dation-reduction catalysts in biological processes. For example, ubiquinones (coenzyme Q)

Pluripotent Carbonyl Reductases of the SDR Superfamily

are constitutive parts of the respiratory chain, and tocopherolquinone (vitamin E) is thought to

protect lipids of biological membranes against lipid peroxidation (Wermuth, 1981).

Due to its capability of daunorubicinol formation, CBR1 has also been named daunorubicin-

pH 6.0 reductase, ALR3 (Flynn and Green, 1993) and xenobiotic ketone reductase with pH

6.0 activity (Ahmed et al., 1978).

Wermuth found that CBR1 metabolized prostaglandin E

and prostaglandin E

, whereas

prostaglandin A

was not reduced at all (Wermuth, 1981). While human CBR1 was shown to

accomplish the reduction of the 9-keto group of prostaglandin E

to form prostaglandin F

the enzyme was later shown to be identical to prostaglandin-9-reductase (Schieber et al.,

1992; Schieber and Ghisla, 1992). However, the metabolic potency of PGE

and PGE

car-

bonyl reduction was only 2% of that observed with menadione. Moreover, the relative veloc-

ity for this reaction was 60-fold lower than for 9,10-phenantrenequinone reduction (Wermuth,

1981). This casts doubt on a role of CBR1 in prostaglandin metabolism. In addition to 9-keto

reduction, CBR1 was shown to catalyze the NADP-dependent oxidation of the hydroxy group

at position 15 of prostaglandins. Therefore, CBR1 has also been named NADP-linked 15-

hydroxy-prostaglandin dehydrogenase. It should be noted that oxo-reduction at position 9 and

hydroxy-dehydrogenation at position 15 are inactivation steps of prostaglandins (cf. chapter

V.1 ; Roles in Steroid and Prostaglandin Metabolism). The fact that the glutathione adduct of

prostaglandin A

is a substrate of CBR1, but free PGA

is not (Wermuth, 1981), might be

explained by the finding that CBR1 has a glutathione binding site in close proximity to the

catalytic centre [reviewed in (Doorn et al., 2004)].

Human CBR1 orthologues in other species

Human CBR1 orthologues with sequence identities of more than 80% to CBR1 have been

identified in several other species such as pig, rabbit, hamster, rat and mouse. The enzyme

from pig testes (PTCR) was the first CBR1 orthologue whose three-dimensional structure has

been solved (Ghosh et al., 1993). PTCR does also exhibit 20-HSD activity towards C

steroids (Tanaka et al., 1992). Whereas in most species the enzyme is distributed in many

tissues, it does not occur in rat liver. Rather, three CBR1 orthologues have been identified in

rat reproductive tissues of both sexes and have been named rat testis (rtCR), gonadotropin-

Pluripotent Carbonyl Reductases of the SDR Superfamily

inducible (iCR) and noninducible (nCR) carbonyl reductase (cf. chapter IV.2.1.5; Rat Carbo-

nyl Reductases (iCR, nCR, rtCR)).

Molecular forms and autocatalytic modification of CBR1

Purification of human brain CBR1 yields three forms which differ in their pI-value, but which

exhibit very similar enzymatic properties (Wermuth, 1981; Inazu et al., 1992b). The three

enzyme forms show isoelectric points of 6.95, 7.85 and 8.5, but no apparent differences in the

amino acid composition (Wermuth, 1981; Nakayama et al., 1985; Inazu et al., 1992b). They

differ from each other only by small structural modifications (see below). Three forms of

CBR1 are also detected on purification from human liver cytosol by gel filtration, ion ex-

change, hydroxy apatite and affinity chromatography (Atalla et al., 2000).

Finally, it was shown that the three isoforms, which show the same catalytic activity towards

menadione (Bohren et al., 1987; Wermuth et al., 1993), result from covalent modifications of

a lysine residue (Krook et al., 1993a; Wermuth et al., 1993; Sciotti et al., 2000). Forrest and

coworkers first reported the occurrence of this modified lysine residue at position 239 of

CBR1, although at that time the nature of the modification was not clarified (Forrest et al.,

1990). This modification is performed by an autocatalytic process involving the formation of

a Schiff base between the -amino group of lysine and 2-oxocarboxylic acids, e.g. pyruvate

and 2-oxoglutarate (Krook et al., 1993c; Wermuth et al., 1993). A covalent adduct is formed

by reduction of the double bond. This autocatalytic modification has never been observed in

other SDRs, nor in another oxidoreductase and is unique to human CBR1 (Sciotti et al.,

2000). A similar modification cannot be found in pig or rabbit carbonyl reductases (see be-

low).

This process is in line with structural data from computer modeling and crystallization data,

which located the modified lysine residue outside of the active site cleft (Krook et al., 1993b;

Tanaka et al., 2005). Sciotti and coworkers suggested a specific structure which is required

for this modification reaction (Sciotti et al., 2000). In human, rat and mouse carbonyl reduc-

tase the region near lysine 239 is highly conserved. For example, rabbit carbonyl reductase,

instead of a lysine, has an asparagine at this position which cannot react with 2-

oxocarboxylates. Mouse carbonyl reductase has not been examined for autocatalytic modifi-

cation by 2-oxocarboxylates. The mouse sequence shows absolute conservation of residues

Pluripotent Carbonyl Reductases of the SDR Superfamily

near lysine 239. This fact suggests that the mouse enzyme can undergo autocatalytic modifi-

cation as well (Sciotti et al., 2000).

The reaction of lysine 239 in carbonyl reductase with 2-oxocarboxylates seems to be physio-

logical, because the three autocatalytically modified forms of human CBR1 are found in un-

treated brain cytosol (Wirth and Wermuth, 1992). The degree of modification of lysine de-

pends on the metabolic state of the cells, because transformed E. coli cells which are incu-

bated in media with glucose show an increased yield of pyruvate-modified CBR1 (Bohren et

al., 1994).

Inhibitors

Enzyme activity can be inhibited by flavonoids, e.g. quercetin and rutin, indomethacin, fu-

rosemide, ethacrynic acid, flufenamic acid and dicoumarol (Wermuth, 1981; Atalla et al.,

2000; Usami et al., 2003). 4-Hydroxy-mercuribenzoate and iodoacetate inactivate the en-

zyme. Neither NADPH nor substrate protect the enzyme from the loss of activity (Wermuth,

1981).

IV.2.1.2 9-Keto-Prostaglandin Reductase and 15-Hydroxy-Prostaglandin Dehydrogenase

Studies on the physiological role of carbonyl reductases indicated an involvement in endoge-

nous prostaglandin metabolism. This field was greatly activated when it turned out that some

prostaglandin dehydrogenases were shown to be SDR enzymes. Two different cytosolic types

of human prostaglandin dehydrogenase have then been characterized in more detail, a mono-

meric NADP(H)-linked enzyme and a dimeric NAD(H)-linked enzyme. Both belong to the

SDR superfamily, have similar conformations regarding their modelled three-dimensional

structure, but they are highly divergent regarding their primary structure, exhibiting identical

residues at only the 20% level (Krook et al., 1993b).

The NADP-dependent enzyme turned out to be identical to carbonyl reductase (EC 1.1.1.184)

which does obviously also act as 9-keto reductase and 15-hydroxy dehydrogenase in pros-

taglandin metabolism (cf. chapter IV.2.1.1; Human Carbonyl Reductase 1 (CBR1)). The

NAD-dependent 15-hydroxy-prostaglandin dehydrogenase is a key enzyme involved in the

biological inactivation of many prostaglandins (Ensor and Tai, 1991). The enzyme is known

Pluripotent Carbonyl Reductases of the SDR Superfamily

to be ubiquitously expressed in several organs in mammals and seems to be downregulated in

cancer tissues (Backlund et al., 2005).

Until today, there is no evidence that NAD-dependent

15-hydroxy-prostaglandin dehydrogenase participates in xenobiotic carbonyl reduction.

IV.2.1.3 Human Carbonyl Reductase 3 (CBR3) and 4 (CBR4)

In their search for genes contributing to Down syndrome, Watanabe and coworkers

(Watanabe et al., 1998) identified a novel carbonyl reductase gene which they named CBR3.

The CBR3 gene is located 62 kb downstream from the original CBR1 gene on human chromo-

some 21q22.2. Comparison of the genomic structure of CBR1 and CBR3 indicated differences

in the introns and surrounding regions but high conservation of the three exons. Coding se-

quence comparisons revealed 77.0 and 84.0% identity on the nucleotide and predicted amino

acid level, respectively, with human CBR1 and, based thereon, CBR3 classified as a mono-

meric NADPH-dependent oxidoreductase.

However, until recently there have been no reports on the catalytic properties of CBR3.

Lakhman and coworkers (Lakhman et al., 2005) were the first to characterize the catalytic

properties of recombinant CBR3 with the prototypical quinone substrate menadione

(Wermuth et al., 1988). While studying the functional significance of a natural allelic variant

of human CBR3, these authors observed a V244M polymorphism that appears common

among different ethnic groups and encodes for CBR3 protein isoforms with distinctive cata-

lytic properties. In detail, blacks showed a higher frequency of the M244 allele than did

whites. In addition, DNA panels from 10 ethnic groups presented a wide range of CBR3

V244M genotype distribution. Comparative three-dimensional analyses based on the structure

of the homologous porcine carbonyl reductase suggested that the V244M substitution is posi-

tioned in a region critical for interactions with the NADP(H) cofactor. These findings support

the notion that CBR3 genetic polymorphisms may impact general CBR1-mediated biotrans-

formations (Lakhman et al., 2005). Further research remains to be conducted to elucidate the

physiological significance of CBR3, as compared to CBR1 and to genetic polymorphisms.

Sequencing of the human genome revealed the existence of a third human isoform of carbonyl

reductase (gene name: CBR4, Swiss Prot accession number: Q8N4T8), based on conserved

domain profiles. The corresponding protein is named carbonic reductase 4 (CBR4). While the

sequences of CBR1 and CBR3 are 72% identical, CBR4 shows only low similarity to CBR1

Pluripotent Carbonyl Reductases of the SDR Superfamily

and CBR3 (23% and 22% identity, respectively). Until now, no data characterizing the en-

zyme regarding function and tissue distribution are available (Strausberg et al., 2002; Ota et

al., 2004).

IV.2.1.4 Chinese Hamster Carbonyl Reductases (CHCR 1-3)

Terada and coworkers (Terada et al., 2001) isolated three different cDNAs encoding carbonyl

reductase in Chinese hamster (CHCR). Comparison of the amino acid composition revealed

that CHCR1 is highly identical to CHCR2 (96%) and to other mammalian carbonyl reduc-

tases (81%). CHCR3 has a high identity to human CBR3 (86%) and a relatively lower

identity to the other CHCRs (76%). Structure prediction of the typical -Rossmann fold

motif of the CHCRs indicated the presence of a typical dinucleotide cofactor- binding motif

which is similar to other SDR enzymes.

CHCR1 and CHCR2 show potent reductase activities towards 4-benzoylpyridine, 4-

nitrobenzaldehyde and pyridine-4-carboxaldehyde. For CHCR2, metyrapone and steroids like

5-androstane-3,17-dione, 5-androstane-3,17-dione and 5-androstane-17-ol-3-one are

better substrates (Terada et al., 2001). CHCR3 has relatively lower activity towards 4-

nitrobenzaldehyde and pyridine-4-carboxaldehyde than CHCR1 and CHCR2.

Terada and coworkers examined the function of the three CHCRs in prostaglandin metabo-

lism (Terada et al., 2003). They found that prostaglandins, e.g. PGE

, failed to act as a sub-

strate for CHCR3, whereas CHCR1 and CHCR2 showed reductase activity towards PGB

and

PGE

. CHCR1 and CHCR2 exhibited dehydrogenase activity towards PGA

, PGB

, PGD

PGE

and PGF

. These findings suggest that both enzymes can catalyze the oxidoreduction

of both the 9- and 15-hydroxy groups of these prostaglandins (Terada et al., 2003). The broad

substrate specificity of CHCR1 suggests that this enzyme is involved in the oxidoreduction of

the 11-hydroxy group of prostaglandins, in addition to 9- and 15-hydroxy groups (Terada et

al., 2003). Additionally, CHCRs show reductase activity towards a variety of androstane and

pregnane steroids, as well as towards benzoylpyridine, daunorubicin and isatin (Terada et al.,

2003).

Pluripotent Carbonyl Reductases of the SDR Superfamily

IV.2.1.5 Rat Carbonyl Reductases (iCR, nCR, rtCR)

Aoki and coworkers (Aoki et al., 1997) isolated two closely related genes encoding an induc-

ible and a noninducible carbonyl reductase in rat ovary. The inducible carbonyl reductase

(iCR) is strongly inducible by pregnant mare serum gonadotropin (PMSG), whereas the non-

inducible carbonyl reductase (nCR) is constitutively expressed. Both carbonyl reductases are

also expressed in rat testis and share a sequence identity of 86% with 277 (iCR) and 276

(nCR) amino acids. The iCR and nCR show sequence homologies of 86% and 80% to human

CBR1, respectively (Aoki et al., 1997).

It is interesting to note that, while human CBR1 is ubiquitously expressed in many tissues, rat

iCR and nCR are only expressed in gonadal and adrenal tissues. There is no carbonyl reduc-

tase expressed in rat liver (Wermuth et al., 1995).

A third carbonyl-reductase-like enzyme is exclusively expressed in rat male and female re-

productive tissues and adrenal glands (rat testis carbonyl reductase, rtCR). Its expression in

the ovary is modulated by gonadotropins and estrogens (Inazu et al., 1992a; Inazu and Satoh,

1994). The amino acid sequence of this enzyme is highly homologous to iCR (98%) and nCR

(86%) from rat reproductive tissues, but with the substitution of some amino acids (5 substitu-

tions to iCR, 37 to nCR) (Aoki et al., 1997).

Recombinant rtCR most efficiently catalyzed the reduction of quinones, e.g. menadione, fol-

lowed by xenobiotic aromatic aldehydes and ketones. Endogenous steroids like dihydrotestos-

terone and 5-androstane-3,17-dione were also accepted as substrates, indicating that rtCR

could be involved in steroid hormone metabolism (Wermuth et al., 1995).

rtCR shows the same length and a 86% positional identity with human CBR1 (Wermuth et

al., 1995). Similar to human CBR1, rtCR can catalyze its own autocatalytic modification by

pyruvate and 2-oxoglutarate (cf. chapter IV.2.1.1; Human Carbonyl Reductase 1 (CBR1)).

Wermuth and coworkers concluded that rtCR and human CBR1 represent species-specific

forms of the same enzyme (Wermuth et al., 1995).

Pluripotent Carbonyl Reductases of the SDR Superfamily

IV.2.1.6 Pig Testicular Carbonyl Reductase (PTCR)

Porcine testicular carbonyl reductase (PTCR) resembles human and rat carbonyl reductases in

that it belongs to the SDR superfamily and catalyzes the NADPH-dependent metabolism of

steroids and prostaglandins, as well as that of xenobiotic aldehydes and ketones (Tanaka et

al., 1995; Nakajin et al., 1997). Because of its ability to reduce the 20-carbonyl group of C

steroids, e.g. the conversion of 17-hydroxyprogesterone to 17/20-dihydroxy-4-pregnen-3-

one, the enzyme is also known as 20-hydroxysteroid-dehydrogenase (Tanaka et al., 1992).

The purified enzyme shows vigorous 3- and 3-HSD activities with 5-androstan-17-ol-3-

one (5-dihydrotestosterone) as substrate. Therefore, the enzyme has sometimes been named

PTCR/3/ (Ohno et al., 1991). Additional names for the enzyme are prostaglandin-E

-9-

reductase and prostaglandin 9-ketoreductase, as well as 15-hydroxyprostaglandin dehydro-

genase (NADP

), because the enzyme catalyzes the NADPH-dependent reduction of pros-

taglandins (Ahmed et al., 1978; Wermuth et al., 1982; Ohara et al., 1995; Forrest and Gon-

zalez, 2000).

PTCR shows a sequence identity to human CBR1 of about 85% (Tanaka et al., 1992) and is

with a score of 80% also highly identical to rtCR (rat testis carbonyl reductase). However,

PTCR lacks the 13 additional amino acid residues at the C-terminus of human and rat car-

bonyl reductases (Tanaka et al., 1995). In contrast to bacterial 3/20-HSD, human 17-HSD

type 1 (Ghosh et al., 1995) and 11-HSD type 1 (Maser et al., 2002; Maser et al., 2003), as

well as Drosophila alcohol dehydrogenase (Benach et al., 1998), which occur as oligomers

(dimers, tetramers), the carbonyl reductases from pig (PTCR), rat (iCR, nCR, rtCR) and hu-

man (CBR1) are monomeric. This monomeric structure is the result of a 41-residue insertion

at a strategic location in front of the conserved Tyr-x-x-x-Lys motif. This insertion describes

an all-helix subdomain that packs against interfacial helices, thus eliminating the four helix

bundle interface conserved in the SDR superfamily and thereby preventing oligomerization

(Ghosh et al., 2001) (cf. chapter III.3.5; Oligomerization and Interfaces). PTCR was the first

known monomeric structure within the SDR superfamily.

Among the SDRs known so far, the carbonyl reductase Sniffer in Drosophila (cf. chapter

3.1.3) shows the greatest similarity to PTCR with regard to crystal structure, apart from the 41

amino acid residue insertion following strand E (Ghosh et al., 2001; Sgraja et al., 2004).

Details

Seiten

Erscheinungsform

Originalausgabe

Jahr

2009

ISBN (eBook)

9783836633918

DOI

10.3239/9783836633918

Dateigröße

10.5 MB

Sprache

Englisch

Institution / Hochschule

Philipps-Universität Marburg – Biologie

Erscheinungsdatum

2009 (August)

Note

1,0

Schlagworte

carbonylreduktase hydroxysteroid-dehydrogenasae proteinstruktur

Autor

Frank Hoffmann (Autor:in)